RE: Additional thoughts on language specification

> 
> > I think to pass this success criterion, each of the following 
> > techniques is in itself sufficient: specifying the language in HTTP 
> > headers or using the meta technique or lang-attribute of the HTML 
> > element or the xml:lang-attribute of the HTML element.
> 
> Exactly.
> 
> > We could recommend specifying the language in the http 
> headers (either 
> > by manipulating the HTTP headers directly or by using the meta
> > technique) combined with the lang + xml:lang attributes of the HTML 
> > element.
> 
> Since the former techniques all are sufficient, I disagree. 
> From my point of view, we should recommend the use of 
> "lang"/"xml:lang" attributes alone. The problem with HTTP 
> headers is that this information is not available in local 
> copies, and the combination of HTTP headers and "lang" 
> attributes bears the risk of contradictory declarations 
> (where HTTP headers win).
> 
> Specifiying a "lang"/"xml:lang" attribute is an easy and 
> sufficient way to declare the language of a document.

Thanks for your comment. I agree that it would be a clear way to present it
to the authors. But I'm not sure it is in sync with what the rest of W3C
recommends since the other references seem to be very keen on the META
technique... 

What would you recommend in the case of a document with two primary
languages? Lang/xml:lang only allow one language. 

If I had to encode such a document, I would use the META technique to
declare the languages of the document as a hole and <div lang="nl"> and <div
lang="fr"> to specify the languages of the sections. 

Yvette Hoitink
Heritas, Enschede, the Netherlands
E-mail: y.p.hoitink@heritas.nl
WWW: http://www.heritas.nl 

Received on Monday, 12 September 2005 13:12:49 UTC